Database enters the new century-shi bole, Zhou aoying, and Zhu yangyong

 

Since the first DBMS system IMS was developed by IBM in 1969, database systems have played an important role in the development of computers. From a hierarchical database system to a network database system, it has become the mainstream relational database system in the database. After the 1980s s, with the improvement of computer hardware technology, computer applications continued to deepen, resulting in many new application fields, such: computer aided design, computer integrated manufacturing, Geographic Information System, office automation, and so on. These new application fields impose new requirements on database systems. There have been research on new databases, such as engineering databases, multimedia databases, fuzzy databases, geographical databases, temporal databases, knowledge bases (also called deductive databases), and object-oriented databases. It is foreseeable that the database system in the future will have the characteristics of deduction, object-oriented and multimedia.

Some research hotspots of today's databases

Relational databases are currently the mainstream products of databases. Although there is still research on relational databases in various international database conferences held every year, with the explosion of digital information, the rapid development of hardware product capabilities and network communication technologies, industrial enterprises have higher and higher requirements on database applications. In addition, the years of database research have laid the foundation for database research hotspots. Let's talk about several aspects below.

1. Object-Oriented Database Technology

Relational databases are well received by users for their complete theoretical basis, simple models, descriptive query languages, and ease of use. With the widespread use of databases, people have put forward requirements for complex applications, and the limitations of relational databases are gradually exposed, such:

(1) The model is too simple to express complicated nesting needs;

(2) The supported data types are limited and cannot support many data structures in programming;

(3) The host language (Programming Language) is separated from the operating language, and the impedance mismatch problem exists.

With the emergence of object-oriented programming languages, such as C ++ and smalltalk, object-oriented databases began to emerge. Object-Oriented Database (OODB for short) is a combination of databases and object-oriented programming languages. These jobs can be roughly divided into three types:

(1) Expand the relational database;

(2) Persistence programming language;

(3) Strict object-oriented database system.

The status quo of the object-oriented database system is as follows:

(1) Persistence programming languages certainly have a lot of interesting results, but they seem undesirable from practicality;

(2) Strict object-oriented database technology. Although there are many achievements, it has not met the original expectations;

(3) Expanding the relational database system. The object-relational database system seems more acceptable to people.

An Object-Oriented Database not only creates classes and objects, but also forms an inheritance level for classes so that sub-classes can inherit attributes and methods from super classes and call methods to access specific objects, but because OODB is a database, it must provide the standard facilities provided by today's relational databases, and relational databases are already widely used, it needs to provide an excessive, allow coexistence with existing RDB and gradually over-develop new products. The proposed standards for object-oriented databases include sql3, odmg93, and odmg95.

2. Knowledge Base (Deductive Database) Technology

Knowledge Base is the combination of artificial intelligence and databases, specifically the product of the combination of logical programming languages and relational databases. The famous computer expert J. D. Ullman (Head of the computer department at Stanford University) Proposed logic as a data model. For the sake of simplicity, the horm clause is used to represent the rule in the form:

Rule: Rule header:-rule body

For the same generation, it can be said that the same parent is the same generation. If the two are the same generation, their children are also the same generation. The rules are as follows:

SG (x, y):-PAR (x, z), par (Y, Z ).

SG (x, y):-PAR (x, XP), par (Y, YP), SG (XP, YP ).

In this example, SG (x, y) indicates that X and Y are the same generation, par (x, y) indicates that Y is the parent of X.

If the apsaradb for redis database stores the PAR (x, y) relationship, you can use the preceding rules to export all the relationships of the same generation from the database.

The relationship stored in the database becomes EDB (Extended Data Base), and the relationship represented by rules is called IDB (inten1_data base ). There are two methods to evaluate the rule: top-down and bottom-up:

The top-down method is to start from the query target and extend to the new target through the rules until it reaches EDB. It is proved that the extension process will be passed under the constraint.

The bottom-up method is based on known EDB facts and uses rules to gradually export IDB facts.

How to improve query efficiency is the key to the knowledge base. Currently, there are many query optimization methods (such as magic set Rewriting Technology), which reduce Repeated Computation and irrelevant tuples.

Knowledge Base research mainly includes query optimization, combination of deduction and object-oriented, parallel distributed knowledge base technology and knowledge base application.

3. Multimedia Database Technology

The term "Multimedia" first appeared at vldb, the Ninth International Database Conference, held in Florence, Italy, on 1983, both the tenth vldb Conference in 1984 and sigmod of the international database conference in 1985 held a special discussion on multimedia databases.

At the end of 1980s, many universities and research institutes developed a large number of multimedia database prototype systems, such as the Minos system developed by the University of Toronto in Canada in 1986 and the historical multimedia database developed by the Grenoble University in France in 1986, in 1988, MCC developed the mini multimedia information manager. Many Italian universities have jointly developed multos, and new prototypes will emerge in the next few years.

Since 1990s, multimedia technology has been greatly developed, as shown below:

(1) audio, image, video, and audio sampling, modulo/data conversion, and storage have been basically solved, providing the possibility for computer processing and application of multimedia information.

(2) the commercialization of computer hardware technologies, especially large-capacity storage devices, and the increasing network bandwidth have laid the hardware foundation for computer processing of multimedia information.

(3) the development and research of various independent media database technologies (such as text library, graphics library, and Image Library) provide the technical basis for multimedia databases.

The emergence of a large number of multimedia applications (such as hypermedia systems and multimedia document systems) promotes the study of multimedia database technology.

However, the original products of today's multimedia database system design the functions and performance of the system based on a specific application. Due to the physical nature of different media, the multimedia application types vary widely, therefore, the main research problems of multimedia are:

◇ Multimedia information modeling provides a unified Multimedia Data Model for multimedia;

◇ Multimedia data indexing, retrieval, access and organization technologies;

◇ Multimedia query language, especially by content;

◇ Multimedia Data Clustering, storage, performance, transmission technology;

◇ Multimedia database system standards.

In addition, parallel database technology, constrained database technology, and active database technology are also the hot topics of today's research. Due to space limitations, we will not talk about them here.

Challenges

In the 21st century, many new application environments have put forward an urgent demand for data management. Example:

◇ The World Wide Web (WWW) environment lacks the support of database technology. With the increasing popularity of WWW applications, processing and management of such a large amount of non-standard data is an important research topic.

◇ Earth Observation System Data and Information System (EOSDIS) is an example of a typical big database. EOS is a group of satellites launched by NASA to collect information about the Earth to support scientists in studying the movements of the atmosphere, oceans, and land. This group of satellites send billions of bytes of information back to the earth, which must be integrated with other data and information. For such a large amount of data, it also needs to meet the various information requests of thousands of science and technology workers and provide an effective data browsing and retrieval mechanism.

◇ In e-commerce applications, heterogeneous information sources need to be integrated organically, which requires high security measures for Distribution Verification and capital flow.

◇ In the medical and insurance information system, a patient's medical records may exist in multiple hospitals. Collaborative medical care requires integration, concurrency, and workflow management.

◇ Computer Integrated Manufacturing (CIM) environment. CIM systems are the key to the survival of enterprises in the current international market competition. In the CIM environment, various design departments, production departments, engineering departments, and administrative departments have their own informatics systems. They are all aggregated to complete a Manufacturing task, with subsystem exchange and collaborative management to achieve efficient manufacturing. These systems involve complex information structures and diverse forms, there are many special requirements for management. For this reason, the international product data exchange standard step/Express has been formulated. At present, many enterprises only follow the standards and standardize their respective data exchange formats. data storage and exchange still use the file format. In this environment, studying database technologies and systems will promote the development of CIM systems.

New Direction of database in the New Century

To adapt to new applications, database research and development cannot be limited to the expansion and improvement of DBMS functions and performance. Further research and development is necessary. Below are some immature research topics:

(1) Schema-free databases: we know that in normal databases, schema is very important. However, in the WWW Network, the data structure is dynamic and it is difficult to apply a fixed pattern. Therefore, it is necessary to study the metadata management facility.

(2) data quality and non-precise query: in the network environment, the quality of different information varies. Data quality includes data timeliness, integrity, and consistency. How to capture and process quality-related source data while obtaining data sources is a problem that must be solved in future database application systems.

(3) scalability and componentization: at present, the database management system is getting bigger and bigger, which is unfavorable for applications with a small amount of processing. The best way is to use database components. You can select different functional components to form a new targeted system based on your needs. Database components facilitate the construction of DBMS modules to improve the scalability.

(4) ease of use of databases: at the 19 vldb Conference held in Dublin, Ireland on 1993, M. stonebraker's "DBMS research at crossroads, Vienna modification opinions" expresses the views of international database experts on the development direction of DBMS. The end user interface is the most important research content, that is to say, DBMS is not easy to use. With the development of computers, there are more and more non-professional users of databases. The main reason why DBMS cannot adapt is that it is not easy to use.

(5) New transaction model: In the past, the transaction processing in DBMS is considered as a small unit to solve the concurrent processing, but it is difficult to process long transactions. In an environment with the rapid development of the network, there may be many concurrent transactions, and the concurrent users are geographically distributed widely. In the past, 2PL protocol was often difficult to adapt to, so new transactions should allow nesting, users should also be allowed to intervene in transaction management.

(6) query optimization: In the future environment, the database type will become very complex and there will be many factors to consider during optimization, such as optimizing such data types, design index methods and query policies, and optimize data accuracy and integrity.

(7) data migration: in a distributed environment, especially in parallel processing, data migration is an important research area. Its optimization should take into account factors such as the communication cost and the cache of intermediate nodes to achieve overall optimization.

(8) Security: in a network environment, allowing various users to access each other is more important to data security. Currently, DBMS security depends largely on the mode. In future applications, data may be in the meta mode. Therefore, we need to study new authorization methods.

(9) Data Mining: Data Mining is currently a rapidly developing research direction. It integrates machine learning, statistical analysis, and database technologies and serves decision makers. In the future, the data volume will be huge. How can we find the rules in the data and provide effective information for decision makers, including the generation, classification, and Sequence Analysis of parallel rules, it is an important research direction with great application value.